Ignorance about the relative importance of events during early epochs currently allows for a rich variety of interpretations of the QSO absorption line systems. A consensus has arisen that most progress can be made when the formation of the absorption systems is treated as an integral part of a general scenario for galaxy formation.
Among the most detailed of such descriptions is the work of Ostriker, Ikeuchi and their co-workers. Ikeuchi and Ostriker (1986) present a unified picture for the formation of ordinary galaxies, young low mass blue compact galaxies and Ly- clouds. They hypothesize the occurrence of explosions (pregalactic objects or expanding voids) at redshifts 5 - 20. The shocks produced by these explosions ionize the intergalactic medium and create shells which are unstable to fragmentation. The large mass fragments collapse to form galaxies, while those with low masses survive as Ly- clouds confined by the pressure of the ambient intergalactic medium.
In this picture there is thus a natural continuity between the Ly- clouds and the dwarf galaxies which might account for the metal line systems.
The chief weakness of this model is that it invokes hypothetical energetic explosions. There is no consensus that these explosions are required to be strong enough to ionize the intergalactic medium (e.g. Donahue and Shull 1987). Rees (1986) notes that the pressure of the intergalactic medium is thus determined in an ad hoc manner, a point which is also made by Vishniac and Bust (1987) who find that the clouds may be confined by ram pressure rather than by the pressure of the ambient intergalactic medium. If the sites of the explosions are young galaxies, then Vishniac and Bust argue that one must either postulate the destruction of clouds in dense environments, or following Salmon and Hogan (1986), require that galaxy clustering be minimal at redshifts 2.5.
Shocks also play a role in many other models. Hogan (1987) has recently discussed a possible origin for the Ly- clouds in the context of the Fall and Rees (1985) theory for the collapse of protogalaxies. Hogan notes that if the Ly- clouds are of relatively low ionization, their contribution to the mass density of the universe is insignificant compared to galaxies (Tytler 1987a). They could then arise in a very minor gas phase occurring during the formation of galaxies. Shocks produced as protogalaxies are assembled from subunits would be observed as Ly- clouds with total column densities in excess of 1017 cm-2, with lower N(HI) occurring in clouds which are moderately ionized. Since the protogalaxies would have sizes of 300 kpc prior to collapse, coherent gas flows might produce Ly- lines which are closely correlated in velocity over lines of sight separated by tens of kpc, as required by the observations of Foltz et al. (1984). This model avoids the need to postulate a medium to confine the Ly- clouds, but it is then unclear whether the redshift evolution of the clouds will match the observations, since this will depend on the unknown details of the evolution of protogalaxies.
Attempts to model long lived Ly- clouds without using pressure support center on the use of cold dark matter. Rees (1986) notes that low mass 'minihaloes' are predicted to exist at the present epoch in the cold dark matter theory for the formation of galaxies. Such minihaloes may be able to trap and stably confine gas. Radii of order 8 kpc are indicated, while the gas density would be low enough to allow a high level of ionization. Ikeuchi and Norman (1987) present further details of this type of model. In common with Rees, they stress that there need be no sharp demarcation between the Ly- and the metal line systems, since collapse and star formation are inevitable in those haloes which have somewhat higher masses than the Ly- clouds.
It will be important to test how well these models can preserve the distinction between the Ly- and metal line system in terms of clustering. If both types of system come from the low mass end of the same spectrum of perturbations, then it may be that the clustering of the metal systems is largely due to hydrodynamic velocities of clouds which arise in the same events which produce the metal enrichment.
Prospects for observational discrimination between these interpretations seem excellent. The ionization of the Ly- clouds should be revealed by the comparison of H and He lines. The frequency of occurrence of low temperature Ly- clouds may be estimated from the numbers of lines with extremely low velocity dispersions. Observations of low redshift systems will be especially exciting. Here we can see whether metal or Ly- systems are associated with dwarf galaxies, we can search for the expected changes in clustering, and extend constraints on the evolution of the clouds up to about 10 billion years.
It is appropriate to end with an analogy drawn from the recent history of extragalactic astronomy. Following their discovery, QSOs were widely regarded as a totally unique new phenomena. We are now aware that in spite of great differences in luminosity, appearance, and space density, the Seyfert galaxies display much of the same physics. A great deal has been learnt from this recognition of underlying similarity. It is suggested that a similar situation applies to the Ly- and the metal line systems seen in QSO spectra.